Re: XML-Query language

[Scherler, Thorsten]

Marc Portier wrote:

nope, not that I know of

I thought so looking into the examples.


hm, you are looking at XMLQuery as a way to query around in some kind of 
document-database.  This will only be performant if there are some 
indexes pre-created for the available documents targetted for the 
queries you are going to launch...
The website I mentioned before is doing it with the help of of a DTD. 
That way the parser "knows" where to look. I think the authors where not 
refering on a db but on "normal" xml-Files. I find that aproache quit 
nice :).
Hence you live more or less in the assumption of an XMLDatabase which is 
more or less the natural habitat for XMLQuery implementations

(although there is a group of people that like the syntax more then XSLT 
and would welcome the FLOWR expressions for single file transformations 
as well. Haven't seen it happen however)

The XPath approach of Xindice for 
example(http://xml.apache.org/xindice/guide-user.html#Querying+the+Database) 
has big disadvantages. Consider something like 
http://javaboutique.internet.com/tutorials/Murachs/index-2.html where 
you can put a SQL-String to search a db (I speak of "total" freedom of 
the Query-String). I didn't get something similar working with XPath.  e.g.:
with (where *expr* is a parameter parsed from a formular):
expr=date[date='Nov 28']
...
<xsl:param name="expr" />
...
<xsl:apply-templates select="{$expr}"/>
This is not working, neither *select="$expr"*, nor *select="'{$expr}'"* 
(or I did something wrong). I came to the conclusion that you can't use 
dynamic XPath expression.

Or do you?

What sizes are we talking about? For really small sets you might 
consider aggregating all into one file and using plain XSLT (probably a 
fair amount of xsl:key usage)
Ok, we talking about log-Files from Mail, FTP, HTTP,... So there will 
become quite big.

But again, this has the sound of pre-indexing all over it.  Depending on 
the variety of queries you need to service you could consider making 
these index files yourself or switch to some XMLDatabase. (or even a 
bread-and-buther RDBMS)
But that will make it necessary to know the index (key) to make a query. 
That is not really user-friendly.
This would mean that you process the separate XML files when they are 
being stored. (e.g. to put parts of the info they hold into the RDBMS)

regards,
-marc=
Thanks for your reply Marc, maybe you have some comments ;-)

King regards
</thorsten>


---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]